[flac-dev] [PATCH] Website comparison + fix IE
Martijn van Beurden
mvanb1 at gmail.com
Fri Jan 11 08:50:11 PST 2013
On 11-01-13 15:26, Miroslav Lichvar wrote:
> Very nice. I'd like to see the comparison page updated too. I've just
> few questions on the pdf. Was the comparison done on a 32 or 64-bit
> system?
All comparisons were done on a 64-bit system. However, I did not see
much deviation from tests in the past, so I guess there are no codecs
that take much advantage of it.
> Doesn't using wine negatively affect the performance?
It might indeed affect performance, but theoretically it shouldn't: Wine
is only an emulation layer for system calls, and as lossless codecs
usually don't need a lot of system calls compared to raw processing
power, this should not affect encoding or decoding time much. Josh did
his comparison on Windows, but on Windows there is no easy way to create
a ramdisk, which is a problem as current CPUs are much to fast for any
(non-SSD) harddisk to keep up. I also can't time CPU-time used, only
real time
> Wouldn't it be better to use a logarithmic scale in the graphs for
> encoding/decoding times and speeds?
Yes, it would give a much clearer and fairer picture and I have created
graphs with logarithmic scales in the past, but I found out that most
people find it very hard to read those scales and are used to the
'distorted' view of the linear scale. Things like 1x or 300x speed are
easier to understand than 1*10^0 and 3 * 10^2, while it is exactly the
same.
To compensate for this I added 'inverse' graphs to the PDF which show
CPU usage instead of times realtime speed, but I think that two graphs
are enough for the "summary" on the website. 4 graphs are too confusing
I think.
More information about the flac-dev
mailing list